A thorough evaluation of flavor characteristics among various retail peanut oil products using multivariate statistical methods

Flavor is a crucial characteristic of peanut oils, exerting a significant impact on their quality and acceptability. However, given the uncertainty in the correlations between flavor attributes and volatile compounds, assessing the flavor quality of commercial peanut oil retail products (PORPs) based on updated analysis techniques remains challenging. In this study, 21 commercial samples were collected for investigating this correlation. From the sensory evaluations, the 16 fragrant PORPs show differences esp. for the roasted (4.25–10), the sweet (0–4.25), and the steamed (0–4.25) attributes while the roasted attribute is dominant for all the products. For volatile profile analysis with updated analytical technology, the fragrant PORPs also show significant difference in terms of volatile profiles and intensities. The total intensity can differ from 16.37 to 66.59 mg/kg. Multivariate statistical methods (MSMs) revealed the significant correlations between volatile compounds and specific sensory attributes. Notably, 1‐hexanol and phenylethyl alcohol demonstrate a strong correlation with the raw peanut flavor attribute. Benzeneacetaldehyde (BA) exhibited a positive correlation with sweet flavor. dl‐Pantolactone has a positive effect on roasted flavor. Hexanal and 3‐octen‐2‐one are highly related to off‐flavor notes. In addition, based on machine learning (ML), three groups of volatile compounds were found to be highly correlated with specific flavor attributes. Notably, a cluster comprising seven compounds, namely 2‐ethenyl‐6‐methylpyrazine, 2‐methyl‐6‐[(E)‐1‐propenyl]pyrazine, benzaldehyde, BA, nonanal, dl‐pantolactone, and 2‐acetopyridine exhibiting a positive correlation with roasted, steamed, and sweet flavor attributes. Furthermore, recombination experiments were conducted to verify the synergistic effect of these seven compounds on flavor performance and agreements were reached.